llama : add option for greedy sampling with probs

Author: ggerganov

Diff for: common/sampling.cpp

@@ -167,9 +167,13 @@ llama_token llama_sampling_sample(
         llama_sample_grammar(ctx_main, &cur_p, ctx_sampling->grammar);
     }
 
-    if (temp <= 0) {
-        // greedy sampling
+    if (temp < 0.0) {
+        // greedy sampling, no probs
         id = llama_sample_token_greedy(ctx_main, &cur_p);
+    } else if (temp == 0.0) {
+        // greedy sampling, with probs
+        llama_sample_softmax(ctx_main, &cur_p);
+        id = cur_p.data[0].id;
     } else {
         if (mirostat == 1) {
             const int mirostat_m = 100;

Diff for: examples/speculative/speculative.cpp

@@ -118,7 +118,7 @@ int main(int argc, char ** argv) {
     std::vector<seq_draft> drafts(n_seq_dft);
 
     params.sparams.grammar.clear(); // the draft samplers will copy the target sampler's grammar
-    params.sparams.temp = std::max(0.01f, params.sparams.temp);
+    params.sparams.temp = 0.0f;
 
     for (int s = 0; s < n_seq_dft; ++s) {
         drafts[s].ctx_sampling = llama_sampling_init(params.sparams);